If you say so. I think it's pretty comical that you guys can even pretend such a system is not one of generalizations, though.
Yes, MBTI gives applicable operations in terms of behaviors and thoughts, but the test itself spends 70 questions asking you directly which of these thoughts and behaviors describe you best. The validity comes from the fact that all the data is provided by the test-taker himself.
The generalization part comes when it takes your personally stated preferences on these external stimuli and makes guesses about how you will respond to other, similar external stimuli in the future. It's behavioralism, and it IS a generalization because it takes a small sampling of your own behavioral preference and then tries to infer information about other situations.
I totally agree that we must look carefully at the generalizations rising from these definitions...as you just said. How does this make MBTI not a system of generalizations?